Q&A: Dr Pearse Keane – AI-powered diagnosis in ophthalmology
Dr Pearse Keane is a Consultant Ophthalmologist at Moorfields Eye Hospital NHS Foundation Trust, Associate Professor at the UCL Institute of Ophthalmology, and Deputy Director and Chief Data Officer for INSIGHT (one of seven data hubs being led by Health Data Research UK).
Dr Keane has recently joined the MedCity Advisory Board. We took a moment to talk to him about his reasons for joining the Advisory Board, and his fascinating day job using artificial intelligence (AI) to identify disease in ophthalmology.
Q: Can you tell us about why you chose to join the MedCity Advisory Board?
I wear a number of different hats in my professional life. I’m a clinician – at Moorfields I specialise in the treatment of some of the most common causes of blindness in the UK, namely age-related macular degeneration and diabetic eye disease. I’m also an academic – at UCL I lead a multi-disciplinary research group that aims to apply artificial intelligence to healthcare. At INSIGHT, we are focused on the safe and responsible use of eye health-related data on a large scale. And I’m also funded by UK Research and Innovation as one of their Future Leader Fellows – this funding is specifically for people who are at the boundary between different disciplines, or between academia and industry.
All this gives a hint that I’m incredibly interested in interdisciplinary working, and breaking down the boundaries between academia, industry and healthcare provision. This is where the ‘magic’ happens and where the most exciting breakthroughs in life sciences are made possible. So, I was excited to be asked to join the Advisory Board, because the MedCity mission to facilitate and convene all the different players in the ecosystem ties in so nicely with my own feelings. Working in collaboration rather than silos is how we can best achieve meaningful patient benefits on a large scale, and that’s what MedCity helps support.
Q: How did your background as a clinician and academic in ophthalmology lead you to working with AI?
Yes, I’m not a computer scientist or engineer but I’ve found myself leading a research group focused on AI! I’ve always been interested in new technologies and innovation, and my previous funding with the National Institute for Health Research involved looking for new innovations in the tech world and applying them to healthcare.
Around 2013/14, when I read about the excitement around how AI and deep learning were reinventing the tech industry, I became very interested in this area, and how it might be applied to ophthalmology, but also to healthcare more broadly. The ‘eureka’ moment came in 2015, when I read about Google’s DeepMind, and discovered that the founders were interested in using AI for real-world benefits. Not only this, but two of the three founders are from London, their HQ was in London, and there were links with UCL too. I approached them, and that quickly led to a meeting that ultimately resulted in a pretty innovative research collaboration.
Q: Tell us more about the collaboration with DeepMind…
In 2016 we signed a research collaboration agreement, which resulted in us sharing anonymised data on around 1.2 million retinol scans.
One of the things we’re proudest of, and that we have been praised for, is that we invested a lot of attention into data protection and privacy, information governance and patient and public involvement. We were very conscious that sharing NHS data of any kind with a private company should be done in an ethical way, that met both national and international standards, and that, ultimately, kept patient benefit front and centre.
Our efforts were rewarded when, in 2018 we published the first results of the collaboration in the journal Nature Medicine. The results showed that the algorithm we developed was as good as world-leading Moorfields experts at assessing eye scans for more than 50 different retinol diseases.
We’re now working with Google Health to make this into a product that can be scaled and used around the world.
Q: What could this breakthrough mean for patients?
We see thousands of patients for optical coherence tomography (OCT) scans every week at Moorfields. The sad reality is that ophthalmologists, both in the UK and around the world, just can’t see patients quickly enough, and, whilst waiting, patients can lose sight, sometimes irreversibly. Early diagnosis and treatment are really essential to being able to save the sight of these patients. Using AI in this way allows for patients with the most sight-threatening disease to be picked up early for treatment.
Ultimately, the vision for the future, and the promise of AI, is that world class expertise can be brought out of the hospital and into the community. So that disease can be picked up early through regular community-based eye tests, but also so that monitoring and regular check-ups could happen at your local optician, rather than with a trip to a busy specialist clinic.
Q: How could you foresee AI being used in other realms of healthcare?
I honestly think that the 2020s are the decade we will finally figure out how to use health data properly, at scale, in a way that protects privacy. The technologies emerging in AI and machine learning are really taking privacy concerns into account. And the changes in digital infrastructure and use of the cloud are opening up vast opportunities for data sharing and linking.
Ophthalmology is perhaps a special case, in that it’s the busiest of all medical specialties for clinic appointments – nearly 10% of all clinic appointments across the NHS are for eyes. And further to that, you have thousands of optometrists across the UK, all feeding information on scans into the NHS system. This leads to a rich source of data in the UK and, in fact at INSIGHT we’ve managed to create the world’s largest bioresource of ophthalmic data. Overall though, I think the NHS has the potential to be world-leading in using data and AI in many other areas of healthcare because of the rich, linked datasets we can potentially create.
Q: This month MedCity released a report on AI and data in London. What is your perception of the London AI/data/life sciences scene?
I have experienced training and working in Ireland, California and London. When I re-visited California shortly before the COVID-19 pandemic, I was struck by how happy I felt to be in London right now. There’s a real confluence of the life sciences industry here. We have amazing universities, major hospitals, and global industry players, all concentrated in one area. I think that’s why I believe London is the most exciting place in the world when it comes to doing AI and clinical work.